Claude Fable 5.1: the 1M-Token Model Explained
Claude Fable 5.1 is the most capable model in Claude Code — 1M-token context, investigates before acting, verifies its own work. Here's when to actually use it.
Evgenii Arsentev · PhDReviewed for accuracy by Evgenii Arsentev, PhD · 2026-09-16
Week 36 brought a new model to Claude Code: Fable 5.1. It's not the default on any plan — you have to switch to it deliberately — and it's not always the right choice. But when you hand it a task that's too big for a single sitting, it handles things Opus would stumble on. This guide explains what it is, when to use it, and how to switch.
What is Claude Fable 5.1?
Fable 5.1 is the most capable model in Claude Code as of September 2026 — Anthropic describes it as built for tasks 'larger than a single sitting'. It has a 1 million token context window, investigates before acting, and verifies its own work more often than smaller models. Unlike Opus or Sonnet, Fable always thinks before answering — you can't turn that off — and it adjusts how hard it thinks based on the effort level you set.
What does a 1 million token context window mean in practice?
A token is roughly 4 characters of text. One million tokens is about 750,000 words — more than War and Peace. In practical terms: you can load an entire large codebase, a full research archive, or months of chat history into a single session, and Fable will hold it all in working memory without forgetting what was at the start. Opus 5 also has 1M tokens on Max, Team, and Enterprise. The difference is Fable's ability to sustain autonomy across that whole space without losing the thread.
When should I use Fable 5.1 instead of Opus or Sonnet?
Use Fable 5.1 when:
- 1The task is genuinely huge — a feature you'd normally split across days, a full refactor, a root-cause investigation across many files.
- 2You want to describe the outcome, not the steps — Fable plans its own path; you don't need to micromanage it.
- 3The problem is ambiguous — debugging an outage, an architecture decision, anything where you'd normally have to give lots of hints.
- 4You've tried Opus and it's losing the thread mid-task — Fable holds longer context more reliably.
- 5You don't need Fable when: it's a quick edit, a simple question, or you're on a usage-credit budget and the task isn't that hard — Sonnet will do.
How to switch to Fable 5.1 in Claude Code
Run /model fable in any interactive session. The fable alias always points to the latest Fable version available to your account — today that's Fable 5.1. In the picker, press Enter to save it as your default model, or s to use it for this session only. To check which model resolved: /model with no argument shows the current selection. Requires Claude Code v2.1.257 or later — run claude update if the model doesn't appear in the picker.
/model fable
Enter saves as default; s = this session only. Behind a corporate gateway that routes to Fable 5, type /model claude-fable-5-1 instead.
Does Fable 5.1 cost more?
It depends on your plan. If the /model picker shows 'Requires usage credits' next to Fable, then yes — Fable calls bill to usage credits on top of your subscription, not against your plan's included usage. On plans where Fable is included, you get it at no extra cost. The picker is honest about this before you switch: if you see the notice, treat Fable as a deliberate upgrade you pay for, not your everyday model.
Don't switch to Fable for normal work — you'll burn through usage credits fast and won't notice the difference on simple tasks. Save it for the sessions where you're about to break a big task into three separate chats. Give it to Fable as one task instead, describe the end result, and let it run. That's where the 1M context and autonomous planning actually pay off.
The model ladder in Claude Code now has four real rungs: Haiku for quick tasks, Sonnet for daily building, Opus for hard reasoning, and Fable for the long hauls. Pick the right gear, not the shiniest one.
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Reading is a good start
Want to actually build this?
Guides explain. The free course transforms — personalized, gamified, and built to get you shipping fast.
Start the free course →